org
DOI: 10.1887/0750306076/b1388v1c8
Optical waveguide theory
G Stewart
From
Handbook of Laser Technology and Applications
Colin E Webb, Julian D C Jones
IOP Publishing Ltd 2005
ISBN: 0750306076, 0750309601
Downloaded on Thu Feb 03 09:03:27 GMT 2005 [193.131.119.40]
Institute of Physics Publishing
Bristol and Philadelphia
Terms and Conditions
Optical waveguide theory 223
A6
Optical waveguide theory
G Stewart
A6.1 Introduction
Over the last 30 years or so, much effort has been devoted to advancing the theory and practice of optical
waveguides. This has been mainly driven by the widespread deployment of optical bre communication
systems where system performance has been dramatically improved over the last three decades. In addition
to the optical bres themselves, a number of optical components, such as diode lasers, couplers and external
modulators, are formed on optical waveguide structures and their development has demanded a thorough
understanding of waveguide theory. On a smaller scale, and particularly since the early 1980s, there has
been a growing research programme on optical sensor technology which has also provided a stimulus for the
development of new types of waveguide devices and the associated theoretical models.
In this chapter, after presenting a brief review of the various types of optical waveguides, we outline the
key principles and parameters which describe and dene the operation of optical waveguides and bres. The
ways in which propagation through optical bres affects the properties of the guided waves are discussed,
including dispersion and non linear effects. Power transfer between propagating waves is essential to the
operation of a number of components and the fundamentals of coupling theory are reviewed. In summary,
the theory given provides the foundation for understanding the detailed operation of a wide variety of optical
components and systems based on optical bre technology.
A6.2 Basic types of optical waveguides
The simplest form of optical waveguide is the three-layer planar or slab guide shown in gure A6.1(a)
consisting of a central guiding layer sandwiched between lower index layers. If the layers on either side are
of equal index, it is known as a symmetric planar guide, otherwise it is an asymmetric guide. The lower layer
(on which the waveguide may be formed in practice) is called the substrate and the top layer the superstrate.
A further distinction arises depending on the index distribution within the guide. If the layers are of uniform
index then the guide is referred to as a step-index structure, whereas if the index varies (usually within the
central layer) it is known as a graded-index guide. As discussed later, the central (waveguiding) layer is
typically a few micrometres in thickness for single-mode operation.
Typical rectangular or two-dimensional (2D) guides are illustrated in gures A6.1(b) and (c). Here
the central (waveguiding) layer is conned to a narrow channel or strip a few micrometres in dimension.
Depending on the fabrication technology employed, the 2D guide may have a ridge structure or be embedded
within a planar substrate. As with planar guides, the 2Dguide is classied as step or graded index in structure.
Optical bre waveguides, shown in gure A6.1(d), (e) and (f ), consist of a circular core surrounded by
a lower-index cladding. Typically, standard single-mode bres have a core/cladding diameter of 9
_
125 m
whereas multi-mode bres have dimensions of 50
_
125 m or 62.5
_
125 m and may be step or graded
224 Optical waveguide theory
Figure A6.1. Various types of optical waveguide: (a) planar, (b) ridge, (c) embedded, (d) single-mode bre, (e)
multi-mode step-index bre and (f ) graded-index bre.
index in construction. In terms of transmission data rates, single-mode bres are superior, followed by the
graded-index multi-mode type.
A6.3 Planar and rectangular guides
A6.3.1 Planar guides
The simple planar guide, with a ray optics approach, provides a useful starting point for understanding the key
properties of optical bres and waveguides in general. If we rst consider a light ray incident on the boundary
between two materials of index n
1
and n
2
, where n
1
< n
2
, as shown in gure A6.2(a), then if exceeds the
critical angle,
c
, given by sin
c
= n
1
/n
2
, total internal reection (TIR) occurs. In TIR, the light actually
penetrates a short distance beyond the boundary, i.e. the eld does not abruptly drop to zero at the boundary
but decays exponentially on the lower index side. This exponential eld is called the evanescent eld and has
an associated penetration (or 1/e) depth. As a result, light which has undergone TIR is phase-shifted from
Planar and rectangular guides 225
Figure A6.2. (a) Expanded view of total internal reection (TIR), (b) light trapped in guide by TIR and (c) transverse
resonance condition for guided modes.
the incident light by an amount 2
1
where
tan
1
=
_
sin
2
si n
2
c
cos
(A6.1)
In equation (A6.1), = 1 for s-polarization (i.e. an electric eld perpendicular to the plane of incidence) and
= n
2
2
/n
2
1
for p-polarized light.
As noted earlier, a planar optical waveguide is formed by a higher index layer sandwiched between
regions of lower index and so light rays may be trapped in the core layer by TIR whenever >
c
as shown in
gure A6.2(b). However, for guiding to occur, a standing-wave pattern must be established across the guide
226 Optical waveguide theory
Table A6.1. Cut-off thickness d
C
for several modes of a slab guide (parameters in text).
Mode TE
0
TM
0
TE
1
TM
1
TE
2
TM
2
d
C
(m) 0.51 0.58 1.81 1.89 3.11 3.19
since the wave must be conned between the boundaries. Put in another way, the round-trip phase change in
the transverse (x-direction, see gure A6.2(c)) must satisfy
_
2
x
2d
_
+2
1
+2
3
= 2m (A6.2)
where m is an integer and
x
is the spacing of the wavefronts in the x-direction,
x
= / cos =
0
/(n
2
cos ).
Hence
k
0
dn
2
cos = m +
1
+
3
(A6.3)
where m = 0, 1, 2, . . . is the mode order, with m = 0 for the fundamental mode and k
0
= 2/
0
.
Since m is an integer, this condition implies that only a discrete set of values of are allowed,
0
,
1
,
2
, . . . ,
m
and each allowed value of corresponds to a certain transverse standing-wave pattern or
guided mode. (Compare standing waves on a string, xed at both ends, giving rise to different modes of
vibration.) The modes which arise from s-polarized rays are transverse-electric (TE) modes because the
electric eld (but not the magnetic eld) is entirely transverse to the propagation direction (z-direction) of
the mode. Similarly p-polarized rays give rise to transverse-magnetic (TM) modes. Hence, in general, a
multi-mode guide will support two classes of modes, TE
0
, TE
1
, TE
2
, . . . , etc and TM
0
, TM
1
, TM
2
, . . . , etc.
Since is polarization-dependent, corresponding TE and TM modes have slightly differing values of and,
hence, propagation constant (this is waveguide birefringence).
A very useful denition for describing the propagation of a guided mode is its effective-index value.
With reference to gure A6.1(c), the phase velocity of the ray in the guide is c/n
2
but since the ray zig-zags
at an angle , the wave-fronts of the guided mode propagate in the z-direction with a phase velocity of
c/(n
2
sin
m
) = c/n
e
where n
e
= n
2
sin
m
is the effective index of the mode. Because >
c
for all
guided modes, the allowed range of n
e
is n
2
> n
e
> (n
1
and n
3
) and the propagation constant of the mode,
m
= k
0
n
e
. With this denition, the eigenvalue equation for the modes (A6.3) can also be written in the
form:
k
0
d
_
n
2
2
n
2
e
= m +
1
+
3
. (A6.4)
The condition n
e
= n
1
or n
e
= n
3
(whichever is the greater) corresponds to the cut-off condition for a mode,
since at that point the ray is travelling at the critical angle at one of the boundaries. This condition may be
used in equation (A6.4) to determine the cut-off thickness (or cut-off wavelength, if the thickness is known)
given the other waveguide parameters, for a particular mode of order m. To illustrate, table A6.1 gives typical
cut-off values calculated from equation (A6.4) for a lm of index n
2
= 1.5 on a quartz substrate (n
3
= 1.45)
with air superstrate (n
1
= 1) at a wavelength of 1 m.
The cut-off thickness for a mode is the minimum thickness which will support that mode. Using the
data in table A6.1 as an example, if the thickness is in the range 0.51 < d < 0.58 m then the guide will
support the TE
0
mode only. Data of this type are used to design waveguides for single-mode operation or for
operation in a selected number of modes, by choosing the appropriate guide dimensions.
At this point, let us consider in some detail the symmetric slab (which provides a very simple planar
model for an optical bre) and introduce a parameter known as the V-number or normalized frequency. The
Planar and rectangular guides 227
Figure A6.3. Universal dispersion curves for a symmetric planar guide.
V-number is a very important parameter for determining the number of modes supported by a guide. By
analogy with an optical bre, the V-number for the planar (slab) guide may be dened as
V
slab
= k
0
_
d
2
_
_
n
2
2
n
2
1
. (A6.5)
(Note that d/2 is the half-width of the planar guide which is analogous to the bre radius a as used in the
denition of the bre V-number in equation (A6.12).)
In addition, it is customary to dene a normalized propagation constant, or b-parameter, which lies in
the range 0 < b < 1, as follows:
b =
n
2
e
n
2
1
n
2
2
n
2
1
(A6.6)
With these denitions, equation (A6.4) can be recast in a form which applies to any symmetric slab guide:
2V
1 b = m +2 tan
1
_
b
1 b
. (A6.7)
Figure A6.3 shows universal bV curves (or dispersion curves) plotted fromequation (A6.7) with = n
2
2
/n
2
1
chosen as 1.2 to show waveguide birefringence. Note that mode cut-off occurs at b = 0, giving the cut-off
value for a mode of order m as V
C
= m/2 and that TE and TM modes have the same cut-off points in a
symmetric guide. Also for single-mode operation, 0 < V < /2, so that in theory the fundamental mode is
always supported in a symmetric guide (compare the asymmetric slab as illustrated in table A6.1 where TE
and TM have different cut off points and the fundamental mode is cut off below a certain value).
Waveguides which are made by diffusion or ion-exchange processes generally have a graded-index core.
In terms of ray optics, we can visualize the ray as being continuously refracted and travelling in a curved path
228 Optical waveguide theory
Figure A6.4. Graded-index slab guide.
through the graded index medium as illustrated in gure A6.4. By applying the same procedure of requiring
the round-trip phase change in the transverse direction to be an integer number of 2, the following equation
may be derived [1] for the modes of a graded-index guide:
k
0
_
d
m
0
_
n
2
(x) n
2
e
dx = m +
1
4
+ (A6.8)
where n(x) is the index prole of the guide, d
m
is the mode depth given by n(d
m
) = n
e
and 2 is the usual
phase shift at the superstrate interface. Note that if the guide is buried into the substrate (e.g. a symmetric
parabolic prole with maximum index below the surface), then 2 = /2. For certain prole forms, such as
a linear or parabolic variation in index, the integral in equation (A6.8) may be evaluated in closed form [1, 2].
So far, all the results discussed are derived on the basis of a simple ray-optics model of a planar
waveguide. A more rigorous method is to obtain the eld solutions by application of Maxwells equations.
The guided modes are assumed to propagate in the z-direction as described by a travelling-wave term of the
form: exp i{t
m
z}. For a planar guide, the transverse-eld distribution is assumed to vary in only the
x-dimension so that the Maxwell wave equation is simplied by the assumption that the
2
/y
2
term is zero.
From the one-dimensional wave equation, one can then obtain the transverse eld component (for example,
E
y
for TE modes) in the form of either a decaying exponential for the cladding regions or as a co-sinusoidal
distribution in the guiding core. The other eld components may then be obtained from E
y
by application
of Maxwells curl equations. When continuity conditions are applied on tangential eld components at the
boundaries of the guide the same eigenvalue equation (A6.4) is obtained for the guided modes. For TE modes
in a step-index planar guide, the TE eld component, E
y
, can thus be obtained in the form:
Substrate (x < d) : E
y
= C
cos
3
cos
1
exp {
3
(x +d)}
Core (d < x < 0) : E
y
=
C
cos
1
cos {k
x
x +
1
} (A6.9)
Superstrate (x > 0) : E
y
= C exp {
1
x}
where + is for even-mode orders and is for odd-mode orders, C is a arbitrary constant, is the decay
constant of the evanescent eld,
i
= k
0
_
n
2
e
n
2
i
and k
x
= k
0
_
n
2
2
n
2
e
. Note that in these equations
the eld is matched at the boundaries x = 0 and x = d and all the elds should be multiplied by the
travelling-wave term: exp i{t
m
z}. The constant C is the eld amplitude at the x = 0 boundary and may
be related to the total power in the guide.
Figure A6.5 shows the eld distribution, E
y
, for the rst few TE modes of an asymmetric step-index
slab guide. Note the cosine distribution of the eld in the core, forming a standing-wave pattern, with
an increasing number of cycles for increasing mode order. The cosine distribution is connected smoothly
(through the continuity conditions as noted earlier) to the evanescent eld tails which penetrate the substrate
and superstrate. The evanescent eld penetration depth, dened by d
p
= 1/ , increases with mode order
Planar and rectangular guides 229
Figure A6.5. Field distribution of modes in a planar step-index slab guide.
and is larger in the substrate for n
3
> n
1
. Each of these eld patterns represent a particular way in which the
light is guided by the structure and, hence, is described as a mode of the guide.
A6.3.2 Two-dimensional guides
Planar structures are generally impractical for guiding light because of diffraction spreading of the beamin the
plane, so 2D or channel guides are required for the manufacture of integrated optic guided wave devices such
as Y-junctions, branching elements, couplers, interferometers and modulators [3]. However, the theoretical
description and analysis of 2D guides is complex and, in general, a rigorous electromagnetic solution is
required. There are some approximate methods that can be applied, mainly based on the extension of planar
waveguide concepts, which we briey consider here.
First of all, Marcatili [4] provided an approximate electromagnetic solution in closed form for the
general structure shown in gure A6.6(a) under the assumption of small index differences and rigorous eld
matching along the sides only (not in the shaded corners where the eld is relatively negligible). Under
these conditions one can regard the light as travelling nearly axially down the guide and consequently the
modes are essentially TEM in character with electric eld polarization along either the x- or y-axis. The eld
distribution in the xy plane (transverse to the propagation direction) may then be constructed in the form
of a product: E(x, y) = E
1
(x)E
2
(y) where E
1
(x) and E
2
(y) have the form of planar-type solutions as in
equation (A6.9). For example, for the y-polarized mode we would write:
in the core: E
y
= C cos(k
x
x +
x
) cos(k
y
y +
y
)
in the region:x > 0 : E
y
= C cos(
x
) cos(k
y
y +
y
) exp{
1
x}
(A6.10)
and so on.
Following on from the Marcatili approach, the mode parameters may be obtained by the effective index
method [1, 5] which makes use of planar mode conditions. To understand this, gure A6.6(b) shows a simple
ray-optics picture of a mode in the 2D guide where the ray undergoes total internal reection in the sequence:
A (bottom) B (left side) C (top) D (right side) A
g
=
_
d
d
_
1
=
_
n
c
+
c
dn
d
_
1
=
c
N
g
(A6.17)
where the group index N
g
is given by
N
g
=
_
n +
dn
d
_
=
_
n
0
dn
d
0
_
. (A6.18)
The transit time (or group delay) for the pulse is therefore t = (L/c)N
g
. If the pulse is generated by a source
with a wavelength spread of = (
2
1
), then the transit time for the pulse will also have a spread
in values because the group index is wavelength dependent. The pulse broadening as a result can thus be
approximated by
t = (t
2
t
1
) =
L
c
{N
g
(
2
) N
g
(
1
)}
=
L
c
dN
g
d
0
. (A6.19)
Using equation (A6.18) for the group index gives the result:
t =
L
c
0
_
2
0
d
2
n
d
2
0
_
. (A6.20)
This expression shows that the pulse broadening depends on the relative spectral linewidth of the source and
the chromatic dispersion coefcient
2
0
(d
2
n/d
2
0
). Note that pulse broadening occurs irrespective of the sign
of t (positive longer wavelengths have longer transit times, negative longer wavelengths have shorter
times). The dispersion is often expressed in terms of a parameter, D, with units of ps nm
1
km
1
through
the denition:
D =
1
c
0
_
2
0
d
2
n
d
2
0
_
=
0
c
d
2
n
d
2
0
. (A6.21)
This denition means that pulse broadening is simply given by: t = DL. Figure A6.16 shows typical
values for the material dispersion parameter D
m
for conventional silica optical bres. Note that the coefcient
is negative for
0
< 1270 nm and positive for
0
> 1270 nm.
Consider now the case of pulse propagation in a waveguide. The propagation constant of a guided
mode is
g
= (2/
0
)n
e
where the effective index, n
e
, is a function of
0
and so equation (A6.20) describes
pulse broadening where the dispersion coefcient is now
2
0
(d
2
n
e
/d
2
0
). This dispersion coefcient may, in
principle, be derived from the characteristic equation for the modes of a guide (for example equation (A6.4)
for planar guides), taking into account the wavelength dependence of the refractive indices.
For this purpose, it is convenient to consider the dispersion in terms of two contributions, namely material
dispersion and waveguide dispersion. As already noted, material dispersion arises from the wavelength
Propagation effects in optical bres 241
Figure A6.16. Material, waveguide and total dispersion for a conventional single-mode bre.
dependence of the refractive indices of the glasses making up the guide, whereas waveguide dispersion arises
fromthe nature of guided modes (the bV curves shown earlier are essentially waveguide dispersion curves).
Consider the most important case, that of dispersion in a single-mode bre [6, 9, 11]. Assuming a
small index difference between core and cladding (as for LP modes), = (n
2
n
1
)/n
1
1, then from
equation (A6.6) the effective index can be written as
n
e
= n(b+1)
and, hence
dn
e
d
0
=
dn
d
0
+n
db
d
0
(A6.22)
where n = n
1
n
2
and it has been assumed that the dispersion of n
1
and n
2
are similar so that is not a
function of
0
.
The dispersion parameter can now be expressed as a sum of material and waveguide dispersion effects
in the form:
D =
0
c
d
2
n
e
d
2
0
0
c
d
2
n
d
2
0
n
0
c
d
2
b
d
2
0
= D
m
+ D
wg
(A6.23)
where D
m
is the material dispersion as given earlier and D
wg
is the waveguide dispersion given by:
D
wg
= n
0
c
d
2
b
d
2
0
=
n
c
0
V
d
2
(bV)
dV
2
(A6.24)
242 Optical waveguide theory
where the approximation V
= k
0
an
(slow mode)
and
a
=
av
a
)L = and the phase difference has reached a value of , the summation of the elds gives
maximum power in guide 2 and minimum in guide 1. Using the values given for the propagation constants,
s
and
a
, the coupling length is, therefore
L =
2
. (A6.26)
Note that, after another coupling length, the phase difference becomes 2 and so maximum power is back in
guide 1. The power is thus cyclically transferred between the guides over their interaction length.
A very useful approximate technique for the theoretical analysis of mode coupling is based on the
concept of weak coupling which leads to the coupled mode equations [3, 1820]. Returning to gure A6.17,
if the coupling between the guides is weak, the (transverse) eld distribution of each guide in isolation will
only be slightly perturbed by the presence of the other guide. Hence, the eld distribution, , of the whole
structure can be approximated by
(x, y, z) A
1
(z)
1
(x, y) exp(i
1
z) + A
2
(z)
2
(x, y) exp(i
2
z) (A6.27)
where
1
and
2
are the normalized transverse eld distributions of guides 1 and 2 in isolation and A
1
(z)
and A
2
(z) indicate that the eld amplitudes vary with distance z along the guides due to the coupling.
Under these assumptions, ,
1
and
2
must all satisfy the wave equation since they all represent
solutions of a guiding structure. When these functions are substituted into the wave equation (with some
approximations which are valid for weak coupling), the following two relationships are obtained [18]:
dA
1
dz
= i
12
A
2
exp(iz)
dA
2
dz
= i
21
A
1
exp(+iz) (A6.28)
where = (
2
1
) and =
12
21
is the coupling coefcient which depends on the overlap between
the elds
1
,
2
of the two guides.
246 Optical waveguide theory
Figure A6.20. Power in each guide for a directional coupler: (a) phase-matched case and (b) mismatched case.
Equations (A6.28) are known as the coupled mode equations and show that variations in the amplitude
in one guide are linked to the amplitude in the other guide through the coupling coefcient. Note that if = 0
(i.e. no interaction between the guides), then the amplitudes in each guide, A
1
and A
2
, remain constant along
the z-direction, as expected.
For the particular case illustrated in gure A6.19 where at z = 0 power, P, is launched into guide 1 with
no power in guide 2, the solution of the coupled mode equations gives the power in each guide as a function
of z:
P
1
(z) = P
_
1
1
1 +
sin
2
z
_
P
2
(z) = P
_
1
1 +
sin
2
z
_
(A6.29)
where
(1 +) and =
_
2
_
2
. Equations (A6.29) reveal the dependence of the coupling length
and the maximum power transferred to guide 2 on the degree of mismatch, , between the guides. Note
that the power fraction transferred is 1/(1 + ) at a coupling length of /2
0
d
e
_
1 +
_
3
k
x
_
2
_ (A6.30)
where:
3
= k
0
_
n
2
e
n
2
3
k
x
= k
0
_
n
2
2
n
2
e
and d
e
= d +
1
1
+
1
3
.
These principles underpin the operation of a variety of integrated and bre optic devices based on direc-
tional coupling. For example, by designing the guides with an appropriate interaction length in relation to the
coupling length, bre or integrated optic couplers with different power-splitting ratios may be manufactured.
Also, since the coupling is stronger at longer wavelengths, as seen by the exp (
3
s) termin equation (A6.30),
wavelength splitters can be designed by choosing an interaction length which corresponds to the coupling
length (or a multiple of it) for
1
but not for
2
, so that maximumcross-coupling occurs at
1
and minimumat
2
. Active devices are made by using directional couplers in electro-optic materials such as lithium niobate.
The index change from the applied voltage is used to match or mismatch the propagation constants of the
two guides, thus forming a modulator or switch.
This discussion has dealt with coupling between modes which are matched or nearly matched in
propagation constants. It is also possible to couple modes which have substantially different propagation
constants if periodic coupling is introduced [22] by, for example, a diffraction grating. This can be seen
from equation (A6.28) by introducing a periodic coupling coefcient of the form:
21
12
= exp (ik
c
z).
Combining this with the exponential terms in equation (A6.28) gives the terms: expi( k
c
) so that
matching now occurs when ( k
c
) = 0 where k
c
= (2/) and is the periodicity of the perturbation.
This phenomenon may cause unwanted coupling between the modes within a waveguide or may be exploited
in certain types of ber optic sensors and narrow-linewidth bre Bragg gratings for use in optical transmission
networks [23, 24].
A6.7 Conclusion
This chapter has outlined the fundamental concepts and principles involved in guiding light in various types of
structures. Knowledge of these principles and the various parameters that have been dened is essential in the
analysis and design of optical waveguides and waveguide-based components. Further detailed information
on bre optic systems, waveguide components and their applications in bre- and integrated-optics may be
found in the list of further reading.
248 Optical waveguide theory
References
[1] Lee D L 1986 Electromagnetic Principles of Integrated Optics (New York: Wiley) pp 11635
[2] Stewart G, Millar C A, Laybourn P J R, Wilkinson CDW and De La Rue R M1977 Planar optical waveguides formed by silver-ion
migration in glass IEEE J. Quantum Electron. QE-13 192200
[3] Syms R and Cozens J 1992 Optical Guided Waves and Devices (London: McGraw-Hill) pp 21749 and 2536
[4] Marcatili E A J 1969 Dielectric rectangular waveguide and directional coupler for integrated optics Bell Syst. Tech. J. 48 2071102
[5] Hocker G B and Burns W K 1977 Mode dispersion in diffused channel waveguides by the effective index method Appl. Opt. 16
11318
[6] Keiser G2000 Optical Fiber Communications 3rd edn (New York: McGraw-Hill) pp 10, 4356, 92103, 10915, 1235, 496500,
50513
[7] Belanger P 1993 Optical Fiber Theory (Singapore: World Scientic) pp 10021, 527
[8] Cherin A H 1987 An Introduction to Optical Fibers (Singapore: McGraw-Hill) pp 85100
[9] Gloge D 1971 Weakly guiding bers Appl. Opt. 10 22528
[10] Kaminow I P 1981 Polarization in optical bers IEEE J. Quantum Electron. QE-17 1522
[11] Senior J M1992 Optical Fiber Communications Principles and Practice (New York: Prentice-Hall) pp 8894, 113, 12130, 9056
[12] Olshansky R and Keck D B 1976 Pulse broadening in graded index optical bres Appl. Opt. 15 48391
[13] Willner A E 1997 Mining the optical bandwidth for a terabit per second IEEE Spectrum 34 3241
[14] Hernday P 2001 PMD posts a speed limit for high speed ber networks Laser Focus World 37 1718
[15] Georges T and Faul J 2000 Soliton transport gives backbones more speed Fibre Systems 4 514
[16] Gordon J P and Haus H A 1986 Random walk of coherently amplied solitons in optical ber transmission Opt. Lett. 11 6657
[17] ForysiakW, Nijhof J HBandDoranNJ 2000Dispersionmanagedsolitons: the keytoterabit per secondoptical ber communication
systems Opt. Photon. News 11 359
[18] Yariv A 1985 Optical Electronics 3rd edn (New York: Holt-Saunders) pp 41349
[19] Ghatak A K and Thyagarajan K 1989 Optical Electronics (Cambridge: Cambridge University Press) pp 44754, 60912
[20] Marcuse D 1971 The coupling of degenerate modes in two parallel dielectric waveguides Bell Syst. Tech. J. 50 1791816
[21] Marcuse D 1987 Directional couplers made of non-identical asymmetric slabs IEEE J. Lightwave Technol. 5 11318
[22] Miller S E 1969 Some theory and applications of periodically coupled waves Bell Syst. Tech. J. 48 2189219
[23] Othonos Aand Kalli K1999 Fibre Bragg Gratings: Fundamentals and Applications in Telecommunications and Sensing (London:
Artech House)
[24] Kashyap R 1999 Fibre Bragg Gratings (San Diego, CA: Academic)
Further reading
Mynbaer D K and Scheiner L L 2001 Fibre-optic Communications Technology ( New Jersey: Prentice-Hall)
Provides a useful starter text for learning about bre optic communications technology. Describes the
operation and characteristics of key bre components with specic, commercial examples. The book is also
suitable for technician training in bre optic systems.
Keiser G 2000 Optical Fiber Communications 3rd edn (New York: McGraw-Hill)
Gives a comprehensive account of all aspects of the design and practice of modern bre communications
systems and networks, including a readable account of bre theory, signal degradation, non-linear effects and
measurements in optical bres.
Senior J M 1992 Optical Fiber Communications Principles and Practice (New York: Prentice-Hall)
Similar in level to Keiser but due for an update.
Belanger P 1993 Optical Fiber Theory (Singapore: World Scientic)
Provides a more advanced and detailed description of the theory of optical bres from the electromagnetic
approach, including the effects of dispersion and nonlinearity on pulse propagation. Theory is well illustrated
by graphical plots and a number of exercises are included.
Cherin A H 1987 An Introduction to Optical Fibers (Singapore: McGraw-Hill)
Somewhat dated but provides a good account of optical ber theory using the electromagnetic approach along
with some practical aspects of optical bre technology.
Further reading 249
Zappe H P 1995 Introduction to Semiconductor Integrated Optics (Norwood, MA: Artech House)
Gives a comprehensive account of semiconductor properties and fabrication technology as applied to inte-
grated optics and describes the construction and operation of a number of important optical components in
semiconductor materials including waveguides, modulators, lasers and detectors.
Lee D L 1986 Electromagnetic Principles of Integrated Optics (New York: Wiley)
As the title suggests, this book presents the electromagnetic theory of planar and rectangular guides and mode
coupling, but also includes optical bers. Examples of integrated optics devices are given as well as the basic
fabrication techniques.
Snyder A W and Love J D 1983 Optical Waveguide Theory (London: Chapman and Hall)
A well-known and classic textbook in the area. Part I of the book gives a full treatment of ray optics in
multimode bres which is not so relevant for modern optical communications systems, but Parts II and III
provide a comprehensive account of the electromagnetic analysis of optical waveguides, including analysis
of bends and perturbations in waveguides and mode coupling between guides.
Syms R and Cozens J 1992 Optical Guided Waves and Devices (London: McGraw-Hill)
Provides a fairly descriptive account of optoelectronic devices including both bre- and integrated-optic
components as well as semiconductor devices with illustrations and applications. The background theory for
understanding waveguide and component operation is also presented.
Hunsperger R G 1995 Integrated Optics: Theory and Technology 4th edn (Berlin: Springer)
Presents the basic theory of waveguides and couplers but concentrates on components and technology and
describes in detail the construction and operation of a number of devices including couplers, modulators and
lasers.